The health effects of later-life
employment
Katey Matthews
ESRC & MRC PhD Student (2010-2013)
CCSR, University of Manchester
Tarani Chandola
Social Statistics and CCSR, University of
Manchester
The proportion of older people in the
workforce is increasing
90%
80%
Employment Rate
70%
60%
50%
1992
40%
2001
30%
2010
20%
10%
0%
16-24
25-49
50-64
Labour Force Survey
65+
Who are older workers?
• Selected for their good health- “healthy
workers effect”
• Heterogeneous occupations- consultants and
cleaners
Most workers aged 65+ are working parttime
% of all in employment by employment
contract
90%
80%
70%
60%
50%
self-employed
40%
working part-time
in a permanent job
30%
20%
10%
0%
50-54
55-59
60-64
Labour Force Survey
65-69
Workers aged 65+ earn less than those in
mid-life
£16
£14
£12
£10
£8
average gross
hourly wage
£6
£4
£2
£0
50-54
55-59
60-64
Labour Force Survey
65-69
Although unemployment among those aged 65+
is low, most are long-term unemployed
60%
50%
unemployment
rate
40%
30%
% of
unemployed
who are
unemployed for
1 year+
20%
10%
0%
50-54 55-59 60-64 65-69
Labour Force Survey
Is working at older ages good for your
health?
Yes
• “Use it or lose” it hypothesis
• Financial security and independence
• Social networks and relationships
No
• Work stress
• “Unfair”
• Less physical activity
Prevalence of suboptimum self-rated health in relation to year of
retirement: GAZEL men who retired at the statutory age of 55 years
High-risk
profile=low grade,
high demands,
and low
satisfaction.
Low-risk profile=high
occupational grade,
low physical and
psychological
demands, and high
job satisfaction.
The health effects of later-life
employment
Katey Matthews, PhD Student
CCSR, University of Manchester
Introduction
• What are the health effects of working beyond
statutory retirement age in the UK?
• Focus on various aspects of wellbeing
–
–
–
–
Mental wellbeing
Cognitive function
Self-rated health
BMI
• Observational study using two datasets:
– English Longitudinal Study of Ageing
– Whitehall II Cohort Study
Background to the topic
• Systematic review provides varied findings:
– 14 studies show a beneficial effect
– 17 studies show a detrimental effect
– 5 studies show no effect
• Meta-analysis also demonstrates high
heterogeneity among results: no real
conclusion can be drawn.
Meta-analysis: results
Depression
Pooled effect (d): -0.171
I2 : 98.5%
Self-rated health
Pooled effect (d): -0.240
I2 : 98.7%
Reitzes 1996
Kremer 1985
Gall 1997
Crowley 1986
Butterworth 2006
Gall 1997
Butterworth 2006
Villamil 2006
Dave 2008
Villamil 2006
Van Solinge 2007
Dave 2008
Zucchelli 2007
Wahrendorf 2008
Coursolle 2009
Zucchelli 2007
McMunn 2009
Roberts 2010
McMunn 2009
Combined
Combined
-2
-1
0
Cohen's d
1
2
-2
-1
0
Cohen's d
1
2
Key problems so far…
• Inclusion of non-comparable study participants
leads to heterogeneity within results:
– Early retirees
– Males and females
– Differing levels of baseline wellbeing
• Subgroup analysis confirms this.
• Need to account for these population differences
in order to discover whether or not later-life
employment does have an effect on wellbeing.
Reducing heterogeneity
• This study uses only comparable individuals:
– Only those over statutory retirement age
– Only those who are definitely classed as ‘employed’ or
‘retired’
– Only those who are employed until reaching retirement
age.
• Propensity score matching used to find the effect of
employment beyond retirement age, as opposed to
retirement.
– “Do differences in wellbeing exist between the groups if all background
characteristics are accounted for?”
Results, conclusions and next steps
• Results of propensity score matching are all
non-significant.
• When comparing later-life workers and retirees
with strictly comparable background
characteristics, there is no significant effect on
wellbeing.
• To what extent do characteristics of
employment affect outcomes?
• Effort, reward and control are all important
considerations.